This paper presents the identification and assessment of the dynamic characteristics of building and bridge structures from the long-term monitoring of seismic response data. Over 40 events of the seismic response data of these structures during the past 10 years were used. To extract the dynamic characteristics of the structures from its seismic response data, the time-domain data-driven subspace identification (SI-DATA) technique is used to extract the system natural frequencies, damping ratios and mode shapes. To enhance the stability of the identified real system poles in using SI-DATA, an algorithm of removing spurious modes is used. Discussion on the identified system dynamic characteristics of the structures under different intensity level of ground excitation is presented. Effect of soil-structural interaction on the identified system natural frequencies and damping ratios by considering either free-field data or basement data as input in subspace identification is also discussed.